Development of a prognostic nomogram based on objective blood test data for patients with advanced cancer undergoing...

Accurate estimation of prognosis can help provide early palliative care to patients. However, few studies have developed nomograms that are totally based on objective blood test parameters. The current study constructed a simple and objective prognostic nomogram and validated the model using advanced cancer patients.

A total of 245 patients were retrospectively analyzed (training sample, n=162; validation sample, n=54), from January 2020 to December 2021. Blood test and demographic data were collated. Cox proportional hazard regression was performed to identify the independent factors, which were built into a nomogram to visualize the probability of patient survival within 30 days. Calibration and discrimination of the model was assessed. The decision curve analysis (DCA) was developed to summarize the performance of the model in supporting decision making.

The median survival was 17.0 [8, 37] days and 21.0 [10, 46] days for the training set and the validation set, respectively. Serum calcium (>2.65 mmol/L), neutrophil count (<2 mmol/L and >7 mmol/L), urea (>7.6 nmol/L), and glutamic oxalacetic transaminase (>40 U/L) were identified and an easily obtained nomogram predicting the 30-day probability of mortality was developed. The nomogram model had adequate discrimination and calibration. The Harrell's concordance index (C-index) of the training set and validation set was 0.69 and 0.71, respectively, while the values of the area under the curve (AUC) of the receiver operating characteristic (ROC) curve were 0.76 and 0.70, respectively.

A simple and objective prognostic nomogram model for predicting the 30-day survival of patients with advanced cancer was developed and validated, with adequate calibration and discrimination. It is expected to guide practical prognosis evaluation in the clinical setting. Further validation is still required in a prospective, multicenter, and large sample study.